N. Gu, S. Watanabe, H. Erhan, H. Haeusler, W. Huang (eds.), Rethinking Comprehensive Design: Speculative Counterculture, Proceedings of the 19th International Conference of the Association of ComputerAided Architectural Design Research in Asia CAADRIA 2014, 000–000. © 2014, The Association for
Computer-Aided Architectural Design Research in Asia (CAADRIA), Kyoto, JAPAN
COGNITIVE EFFECTS OF USING PARAMETRIC
MODELING BY PRACTICING ARCHITECTS:
A PRELIMINARY STUDY
RONGRONG YU,1 JOHN GERO,2 and NING GU3
1,3
University of Newcastle, Newcastle, Australia
rongrong.yu, [email protected]
2
Krasnow Institute for Advanced Study, George Mason University,
Fairfax, USA and
University of North Carolina at Charlotte, USA
[email protected]
Abstract. This paper presents the results of a protocol study which
explores the cognitive behaviour of eight practicing architects while
they used a geometric modeller (Rhino) with a parametric modeller
(Grasshopper) as they designed. The protocol videos collected were
transcribed, segmented and coded using the FBS ontology as the coding scheme. This resulted in each protocol being transformed from a
qualitative video into a sequence of symbols from the FBS ontology
and further divided into design knowledge and rule algorithm classes.
The sequence of symbols forms the foundation on which quantitative
representations of cognitive behaviour can be constructed and compared. Results of the relative cognitive effort expended on design
knowledge and rule algorithm classes, through an articulation of the
cognitive design issues, have been compared and discussed. These results provide insight into the use of parametric modellers by architects.
Keywords. Design cognition; parametric modelling; protocol studies
1. Introduction
Design media have a significant effect on designers’ thinking processes
(Chen 2001; Mitchell 2003). With the increasing use of parametric modelling by practicing architects it is useful to determine what effects the use of
such tools have on their cognitive behaviour. According to Kolarevic (2003),
the change in designing associated with parametricism is characterized by a
2
R. YU, J. GERO AND N. GU
rejection of the static solutions offered in conventional design systems and
the adoption of intelligent systems. Previous studies potentially support this
view, arguing that parametric tools can advance design processes in a variety
of ways (Qian, Chen et al. 2007; Schnabel 2007; Abdelmohsen and Do
2009; Iordanova, Tidafi et al. 2009). However, there is a lack of empirical
evidence supporting an understanding of designers’ behaviour in the parametric design environment (PDE).
In order to examine the way designers think and act in a PDE, this paper
presents the results of a cognitive study in which designers were asked to
complete an architectural design task using parametric modelling tools. Protocol analysis (Ericsson and Simon 1993; Gero and Mc Neill 1998) was then
employed to study the designers’ behaviour. From the analysis, the relative
cognitive effort expended on two classes of design activities – design
knowledge and rule algorithm – through an articulation of the cognitive design issues is presented and discussed.
2. BACKGROUND
Parametric design is a dynamic, rule-based process controlled by variations
and parameters in which multiple design solutions can be developed in parallel. According to Woodbury (2010), it supports the creation, management
and organization of complex digital design models. By changing the parameters of an object, particular instances can be altered or created from a potentially infinite range of possibilities (Kolarevic 2003). The term “parameters”
means factors which determine a series of variations. In architecture, parameters are usually defined as building parameters or environmental factors.
Previous studies on designers’ behaviours in PDEs suggest that parametric tools advance design processes in a variety of ways. For instance, there is
evidence that the generation of ideas is positively influenced in a PDE. Particularly, in Iordanova et al.’s (2009) experiment on generative methods,
ideas were shown to be generated rapidly while they also emerge simultaneously as variations. Schnabel (2007) showed that PDEs are beneficial for
generating unpredicted events and for accommodating changes. However,
researchers have typically studied design behaviour in PDEs by observing
students’ actions within PDEs in design studios or workshops rather than
practicing architects. This empirical gap is addressed in the present study by
adopting the method of protocol analysis in studying practicing architects.
Protocol analysis is a method used for in-depth or detailed analysis of designers’ cognitive activities. Lee et al (2012) have demonstrated the use of
protocol analysis to evaluate creativity in the PDE. Using the same method,
Chien and Yeh (2012) explored “unexpected outcomes” in the PDE. Results
COGNITIVE EFFECTS OF USING PARAMETRIC MODELING
3
of these studies both confirm the viability of the method, and also suggest
that in some conditions PDEs can potentially enhance creativity.
2.1. PROTOCOL STUDIES USING FBS ONTOLOGY
Protocol analysis is a method for turning qualitative verbal and gestural utterances into data (Ericsson and Simon 1993; Gero and Mc Neill 1998). It
has been used extensively in design research to develop an understanding of
design cognition (Suwa and Tversky 1997; Atman, Chimka et al. 1999; Kan
and Gero 2008). In design research a common coding scheme is the one
based on the Function-Behaviour-Structure (FBS) ontology (Gero 1990).
The FBS ontology, Figure 1, contains three classes of ontological concepts: Function (F), Behaviour (B) and Structure (S). Function represents the
design intentions or purposes; behaviour represents how the structure of an
artefact achieves its functions; and structure represents the components that
make up an artefact and their relationships. Figure 1 shows the eight numbered design processes that flow from this—formulation, analysis, evaluation, synthesis, documentation, reformulation-1, -2, and -3.
Figure 1. FBS ontology (after Gero and Kannengiesser 2004)
2.2. TWO CLASSES OF DESIGN ACTIVITIES
The ways in which parametric design is used by architects are not well understood which is why Sanguinetti and Kraus argue that parametric design
“requires a deeper understanding of how it can support our intentions as architects” (Sanguinetti and Kraus 2011, p. 47). Compared to traditional design environments, in a PDE designers not only design by applying design
knowledge, but they also define rules and their logical relationships, using
parameters. Thus, in a typical parametric design process, there are two classes of cognitive design activities: the design knowledge and the rule algorithm. In the design knowledge class, architects make use of their design
knowledge including, for example, how to adapt a building to the site, how
4
R. YU, J. GERO AND N. GU
to shape the way people use the building, and how to satisfy the requirements of their clients. In the rule algorithm class, they apply design
knowledge through the operations of the parametric design tools, including
defining the rules and their logical relationships, choosing the parameters
suitable for a particular purpose and importing external data into the proposed rules. During the design process, designers progress by applying design knowledge directly and in some parts they apply design knowledge indirectly by defining rules and their logical relationships, and this is known as
parameterization. In order to capture the processes involved in designing in
PDEs, the main class of variables from the original FBS ontology that have
been used are R, F, Be, Bs, and S. Each variable is then further decomposed
into the two classes of design activities: the design knowledge class, denoted
by the superscript K, and the rule algorithm class, denoted by the superscript
R, Figure 2. For instance, when designers set up algorithm goals or think
about the way to achieve those goals in the rule algorithm space, the segments will be coded as BeR.
Figure 2. Applying FBS ontology in the PDE.
3. EXPERIMENT SETTING
The study reported in this paper is focused on the work of eight designers,
who are all professional architects with an average of eight years of experience, and with no less than two years of experience using Rhino and Grasshopper. In the experiment, each designer was required to complete a defined
architectural design task using a parametric modeller.
During the experiment, both designers’ activities and their verbalization
were video-recorded by a screen capture program and the recorded data subsequently used for protocol analysis. The design environment is Rhino and
Grasshopper, a typical PDE. Designers were given 40 minutes for the design
session, although most needed slightly more time to complete the design.
The design task was a conceptual design for a commercial building containing certain specific functions and located on a pre-modelled site which was
provided to each designer. During the experiment, designers were not allowed to sketch manually so that almost all their actions occurred on the
computer to ensure that the design environment was purely within the PDE.
All of these controls were put in place to ensure that any variables which
COGNITIVE EFFECTS OF USING PARAMETRIC MODELING
5
could potentially bias the study were minimized. This design environment is
close to a real-world design environment at this stage of the design.
4. RESULTS
4.1. GENERAL RESULTS
In order to increase the robustness of the protocol coding process, two
rounds of segmentation and coding were conducted with an interval of two
weeks between them. Thereafter an arbitration session was carried out to
produce the final protocol from the combination of the two rounds. The percentage of agreement between the two rounds was 83.4% [SD=5.7%], and
between the individual rounds and the final arbitrated results was 91.5%
[SD=3.1%]. These percentages are indicative of the methodological reliability of the coding process and results. From analysing the eight protocols, the
average overall numbers of segments—a segment is the division of a protocol into individual units, in the current research, the division was based on
FBS meaning —were 244 [SD=29.7]. The mean time spent on the design
session was 48.4 mins [SD=7.4] and over 92.2% of all segments were coded
using the FBS model. Non-coded segments include those concerned with
communication, software management, etc. The quantitative anlysis of FBS
design issue occurrences are shown in Table 1. The occurrences of design
issues were normalized by turning them into percentages of the total number
of design issues in each protocol. The results indicate that designers spent
most of their cognitve effort on the structure-related design issues which
consist of SK (21.56%), SR (20.49%), BsK (23.55%) and BsR (5.18%).
Although designers show individual differences in their design strategies and
habits, by exploring the repeated patterns across their behaviours,
characteristics of parametric design will be identified.
Table 1. Design issue analysis
R! (%)
F ! (%)
Be! (%)
Be! (%)
Bs ! (%)
Bs ! (%)
S ! (%)
S ! (%)
Mean
1.52
5.09
11.08
11.59
23.55
5.18
21.56
20.49
SD
0.72
2.81
7.96
6.12
5.09
2.80
9.37
10.18
4.2. RELATIVE EFFORT ON THE TWO CLASSES
The protocol’s coding scheme codes six ontological design issues, where a
design issue maps onto a cognitive activation. Since the length of each design session varies the number of segments in session is first normalized to
be 100. This then makes each design session the same length of 100 normalized segments. In order to be able to compare the relative cognitive efforts,
6
R. YU, J. GERO AND N. GU
the cognitive activations associated with design knowledge and those associated with rule algorithm are separated and aggregated across all eight designers.
Function (F) and Requirement (R) in the rule algorithm class were coded
as zero in this study. The three ontological design issues of expected behaviour (Be), structure (S) and behaviour from structure (Bs) are the most frequently occurring issues and measurements are made using those three. At
each normalized segment the relative percentages of cognitive effort on two
classes of design activities – design knowledge and rule algorithm – are calculated and plotted. The resulting graphs provide a qualitative overview of
the locus of cognitive effort distribution between design knowledge and rule
algorithm, in addition to the quantitative values used to produce the graphs.
4.2.1. Overall relative effort on the two classes
The overall relative effort on two classes of designers’ cognitive activities –
design knowledge and rule algorithm is illustrated in Figure 3. The vertical
axis represents the average value of relative effort of eight protocols. The
horizontal axis is the normalized segment numbers.
From the overall distribution of cognitive effort shown in Figure 3, we
can see that initially the cognitive effort on design knowledge dominates that
on rule algorithm. However, as the design session proceeds, the cognitive
effort on design knowledge drops from 100% to approximately 60% of the
total in a shape that looks like a decay curve. In parallel, as the design session proceeds, the cognitive effort on rule algorithm increases from 0% to
approximately 40% of the total in a shape that looks like an excitation curve.
Therefore, we can infer that in parametric design process, designers still expend most effort on design knowledge; parametric scripting is mainly used
to support their intention of generating models. Designers started with considering design knowledge-related issues, such as building functions; as the
design proceeded, they gradually spent more of their cognitive effort on parametric scripting. In the following sections, we articulate these results with
further detail about the cognitive behaviour related to Bs, S and Be, and
draw implications from them.
4.2.2. Relative effort on Be
Expected behaviour (Be) means that “designers use theory or experience to
speculate what effect could fulfil a purpose before a specific structure is proposed” (Jiang 2012, p 36-37). Applying it in rule algorithm related activities
(BeR) it means that designers set up algorithm goals or think about the way
to achieve those goals.
COGNITIVE EFFECTS OF USING PARAMETRIC MODELING
7
Figure 3. Overall relative cognitive effort
The relative cognitive effort expended on expected behaviour (Be) of the
two classes during the parametric design process is presented in Figure 4. In
terms of expected behaviour (Be), the relative effort expended on design
knowledge class (BeK) is higher at the beginning and then decreases across
the design session; while that on rule algorithm class (BeR) rises toward the
end of design session. At the end of the design session, designers’ cognitive
effort expended on the rule algorithm exceeds that of design knowledge.
From this we can infer that later design in the design session, designers tend
to focus more on setting rule algorithm goals and exploring ways to achieve
those goals.
Figure 4. Relative cognitive effort expended on Be
4.2.3. Relative effort on Bs
Structure behaviour (Bs) refers to the behaviour derived from the structure:
for the design knowledge class, Bs ! represents evaluation of existing geometry/structure; while for the rule algorithm class, Bs ! means evaluating the
structure of the rule algorithm.
8
R. YU, J. GERO AND N. GU
The relative cognitive effort expended on structure behaviour (Bs) of the
two classes during the parametric design process is presented in Figure 5.
Designers expended noticeably more cognitive effort on the design
knowledge class is than on rule algorithm class during the whole design session when focusing on Bs. Design knowledge-related activities decrease
from 100% to approximately 70% during the first third of design sessions
and then remain unchanged between 70%-80%; the rule algorithm-related
activities increase from 0% to 30% during the first third of the design sessions and then remain unchanged between 20%-30%. From the results in
Figure 5, we can see that in terms of Bs, rule algorithm related activities
(Bs ! ) do not commence at the beginning of the design sessions. One of the
possible reasons is that designers only examine rule setting after their rule
concepts achieve a certain degree of maturity.
Figure 5. Relative cognitive effort expended on Bs
4.2.4. Relative effort on S
Structure (S) variables describe “the components of the object and their relationships, which mean “what it is”.” (Gero and Kannengiesser 2004, p 374).
In the design knowledge class, structure (S ! ) refers to the elements or relationships of the geometries; while in the rule algorithm class (S ! ), it is defined as the structure of the rule algorithm – the components of rules and
their relationships for parameterization.
The relative cognitive effort expended on structure (S) for the two classes
during parametric design process is shown in Figure 6. From the results in
Figure 6, we can see that design knowledge-related activities dominate the
design process during the early and mid-stages of the design sessions and
then converge to a level similar to that of rule algorithm activities. The cognitive effort expended on S ! decreases from 100% at the beginning of the
sessions to approximately 50%; while rule algorithm-related activities increase from 0% to approximately 50%.
COGNITIVE EFFECTS OF USING PARAMETRIC MODELING
9
Figure 6. Relative cognitive effort expended on S
5. CONCLUSION
This paper presents the results of a cognitive study based on an experiment
with eight professional architects using parametric design tools. From protocol analysis, designers’ relative cognitive effort expended on design
knowledge and rule algorithm class have been explored. Results of this study
suggest that the division of these two design classes is useful in understanding designers’ behaviour in the PDE. The results indicate that the design
knowledge-related activities dominate the parametric design process for all
cognitive issues.
The implications of these results, if they are found to be generalizable, is
that practicing architects with experience in using parametric design tools
make use of those tools very early in a design session and make increasing
use of them as the design session proceeds. This implies that the designer is
substituting rule algorithms for design knowledge.
These preliminary findings have implications for architectural education
in the digital design age. Architectural education should include how design
knowledge is captured in rule algorithms and in parametric design in general
rather than focusing on the structure of rule algorithms only. This opens up
ways of encoding design patterns that can form the basis of reusable rules
that allow a designer to develop a style of designing and through the rule parameterization a style of designs. Each design generated through the use of
these patterns but parameterized individually is unique and responds to each
unique program but forms part of an overall style associated with an individual designer or design team.
10
R. YU, J. GERO AND N. GU
References
Abdelmohsen, S. and E. Y.-L. Do: 2009, Analyzing the significance of problem solving
expertise and computational tool proficiency in design ideation, CaadFutures2009,
Montréal, 273-287.
Atman, C. J., J. R. Chimka, K. M. Bursic and H. L. Nachtmann: 1999, A comparison of
freshman and senior engineering design processes, Design Studies, 20(2), 131-152.
Chen, S.-C.: 2001, The Role of Design Creativity in Computer Media, eCAADe2001,
Helsinki, Finland, 226-231.
Chien, S.-F. and Y.-T. Yeh: 2012, On Creativity and Parametric Design---A preliminary
study of designer’s behaviour when employing parametricdesign tools, eCAADe2012, Czech
Republic, 245-253.
Ericsson, K. A. and H. A. Simon (ed.): 1993, Protocol analysis : verbal reports as data, MIT
Press, Mass.
Gero, J. and U. Kannengiesser: 2004, The situated function-behaviour-structure framework,
Design Studies, 25(4), 373-391.
Gero, J. S.: 1990, Design prototypes: a knowledge representation schema for design, AI
Magazine, 11(4), 26–36.
Gero, J. S. and T. Mc Neill: 1998, An approach to the analysis of design protocols, Design
Studies, 19(1), 21-61.
Iordanova, I., T. Tidafi, M. Guité, G. De Paoli and J. Lachapelle: 2009, Parametric methods
of exploration and creativity during architectural design: A Case study in the design studio,
CAADFutures2009, 423-439.
Jiang, H.: 2012, Understanding senior design students' product conceptual design activities-a comparison between industrial and engineering design students, National Universigy of
Singapore. Doctor of philosophy.
Kan, J. W. T. and J. S. Gero: 2008, Do computer mediated tools affect team design
creativity?, CAADRIA2008, Chiang Mai, Thailand, 263-270.
Kolarevic, B. (ed.): 2003, Architecture in the digital age : design and manufacturing, Spon
Press, New York, NY.
Lee, J. H., N. Gu, J. Jupp and S. Sherratt: 2012, Evaluating Creativity in Parametric Design
Processes and Products: A Pilot Study, DCC2012, Texas.
Mitchell, W. J. (ed.): 2003, Beyond Productivity: Information Technology, Innovation, and
Creativity, The national academies press, Washington, D.C.
Qian, C. Z., V. Y. Chen and R. F. Woodbury: 2007, Participant Observation Can Discover
Design Patterns in Parametric Modeling, ACADIA2007, Halifax, 230-241.
Sanguinetti, P. and C. Kraus: 2011, Thinking in Parametric Phenomenology Parametricism
(SPC) ACADIA Regional 2011 Conference, Lincoln, 39-48.
Schnabel, M. A.: 2007, Parametric Designing in Architecture, CaadFutures2007, Sydney,
237-250.
Suwa, M. and B. Tversky: 1997, What do architects and students perceive in their design
sketches? A protocol analysis, Design Studies, 18(4), 385-403.
Woodbury, R. (ed.): 2010, Elements of Parametric Design, Routledge, New York.